Gene Therapy (1998) 5, 755–760  1998 Stockton Press All rights reserved 0969-7128/98 $12.00 http://www.stockton-press.co.uk/gt A novel method for the direct quantification of gene transfer into cells using PCR in situ

´ C Catzavelos1, C Ruedy1, AK Stewart2 and I Dube1,2 1Department of Laboratory Medicine, Sunnybrook Health Science Centre; and 2Oncology Gene Therapy Program, The Toronto Hospital, Toronto, Ontario, Canada

There are several limitations to current methods for the readily identified by fluorescence microscopy and a high detection of target genes following gene transfer. We sensitivity, specificity and correlation coefficient were dem- report a novel PCR in situ procedure which overcomes onstrated in mixing experiments using varying proportions many of these and permits the direct quantification of gene of known provirus positive and negative cells. The method transfer in individual cells. PCR amplification of a proviral was applied successfully to identify low numbers of gene- specific nucleotide sequence in target cells was followed modified hematopoietic cells in clinical specimens in a trial by in situ hybridization using fluorescent probes com- of retrovirus-mediated gene transfer into blood forming plementary to different regions of the amplicon. Many of stem cells. This approach is simple and reliable, has the the problems previously encountered using in situ PCR, potential for use in a variety of gene therapy applications particularly the generation of false positive results and and may become the method of choice for the assessment extracellular leakage of PCR products, were overcome by of gene transfer efficacy. modifications of existing protocols. Positive cells were

Keywords: in situ PCR; retrovirus-mediated gene transfer; hematopoiesis

Introduction addition, marker genes may induce changes in the expression of endogenous genes which may alter a wide Gene therapy offers an opportunity to translate advances variety of cellular functions.12 in molecular medicine for the direct benefit of the patient. In situ PCR (ISPCR) permits the intracellular detection At this point in the development of genetic therapies, sig- of PCR products and has the potential to document the nificant emphasis remains on the development of safe proportion of cells modified by gene transfer. This tech- 1,2 and efficient gene delivery systems. Using current tech- nique has been used extensively for viral detection,13–15 niques, the proportion of cells bearing the transferred as well as to detect low abundance mRNA16 and single gene is generally less than 5% of total cells present in copy DNA sequences in situ.17 A number of technical 3 patient material obtained after gene transfer. Thus, sensi- difficulties have been encountered using this method tive and specific detection methods are required for however, the most important being generation of false monitoring gene marked cells in vivo. There are several positive results and leakage of PCR product from target limitations to the current methods used to detect target cells.18–21 4–6 genes. For example, PCR does not permit reliable We report the development of a simple and reliable identification of the transferred gene in specific cells or PCR in situ approach which overcomes many of these quantification of transfer efficacy and Southern analysis limitations. Application of this procedure in a clinical is insufficiently sensitive to detect single copy genes in trial of stem cell marking using a neor construct demon- small cell populations. Fluorescence in situ hybridization strated its utility by identifying gene marked hematopo- 7 (FISH) has been used to assess gene transfer but in most ietic cells. This method has the potential to be applied in gene therapy applications, the target size is usually less a number of gene therapy settings in which the evalu- than 10 kb making the procedure technically difficult, as ation of single copy sequences in individual cells is well as creating the requirement for sophisticated equip- required, and may complement existing technologies. ment for analysis and interpretation. Bi-cistronic con- structs incorporating a marker gene, such as B-galacto- sidase8 or green fluorescent protein9 may be used for the Results detection of gene-modified cells, but the presence of an A Jurkat cell line with and without a stable neor integrant additional exogenous gene may affect expression of the was used to develop the method.22,23 Neor positive and 10 11 therapeutic gene or elicit a host immune response. In negative cells were mixed in equal proportion and slides were prepared by dropping cell suspensions pretreated by exposure to hypotonic buffer and fixed in methanol- r Correspondence: C Catzavelos, Sunnybrook Health Science Centre, Room acetic acid on to glass slides. Several neo specific primer E-433, 2075 Bayview Avenue, Toronto, Ontario, Canada M4N 3M5 pairs were evaluated by solution phase PCR and the pair Received 31 July 1997; accepted 13 January 1998 with the most efficient amplification was selected. The Direct quantification of gene transfer with ISPCR C Catzavelos et al 756 PCR product generated was 432 bp. Initially, a protocol samples from individuals enrolled in a clinical trial of involving a single PCR reaction with direct incorporation stem cell gene marking who were undergoing autologous of fluorescein-dUTP into the PCR reaction product was bone marrow transplantation for chemotherapy resistant used. Following a ‘hot start’ procedure to reduce mis- multiple myeloma.24 Post-infusion blood and marrow priming,20 the slides were subjected to 40 PCR cycles. samples as well as PHA/IL2 and EBV-stimulated Thereafter they were washed once for 10 min in 2 × SSC lymphoid cultures were evaluated. Neor-positive cells at 55°C and visualized using fluorescence microscopy. were consistently detected in both peripheral blood and Nonspecific PCR product was detected in the majority of bone marrow samples. There were no ambiguous cells, cells and extensive extracellular leakage was observed. In background or evidence of extracellular leakage.25,26 subsequent experiments, the amount of fluorescein-dUTP in the reaction mixture was decreased, the number of PCR cycles was reduced and the stringency of the post- Discussion hybridization wash increased but nonspecific product formation and extracellular leakage persisted. DNA ISPCR offers a means to detect single copy To address these problems, a two-step approach was sequences in individual cells. There are two approaches used involving an initial PCR reaction with unlabeled in current use. The direct method involves a single in situ primers and subsequent detection by in situ hybridization PCR reaction using labeled nucleotides, while the indirect using two nonoverlapping, fluorescein labeled probes protocol consists of an initial PCR reaction without la- spanning the amplicon. Experiments using 40 amplifi- beled nucleotides, followed by in situ hybridization using cation cycles showed PCR product in the expected per- labeled probes to the PCR products generated. A number centage of neor-positive cells, but extensive extracellular of technical problems have been encountered using this leakage remained a problem. This was markedly reduced technique, the most significant being the generation of by a decrease in the number of PCR cycles from 40 to 15. false positive results.18–21 This has been more commonly An initial post-hybridization wash stringency of 2 × SSC associated with the direct method due to the formation of at 55°C was used but considerable background was nonspecific PCR products from mispriming or products observed. This was eliminated by increasing the post- generated from nonspecific extension of single strand hybridization wash temperature to 62°C (Figure 1a). DNA nicks (end-labeling artifact).18 This creates the Negative controls included pretreatment of slides with requirement for multiple, stringent controls and DNase as well as use of the reaction mixture without Taq additional validation of results particularly when using polymerase or without primers. No positive cells were the direct method. We have attempted both approaches observed (Figure 1b). RNase pretreatment had no effect using multiple controls and shown the indirect method on the results obtained. to be more reliable. This reduces the problem of false As an additional control and to evaluate sensitivity, positive results by allowing visualization of only specific specificity and the correlation coefficient of this method, amplicons in contrast to those generated by mispriming mixing experiments were performed. Provirus-positive or end-labeling. In the direct approach, all PCR products Jurkatt cells were diluted with negative cells in varying are detected due to direct incorporation of labeled proportion (0–100%), subjected to ISPCR as outlined nucleotides. above and scored in a blinded fashion. The experimental Another significant obstacle has been leakage of PCR data is shown in Table 1. The calculated sensitivity and product from target cells.18,19,21 In some studies, false specificity was 96 and 93.7%, respectively. The Pearson positive results have occurred as a result of diffusion of correlation coefficient comparing actual with observed amplicons into nontarget cells.26 In our experiments, this score was 0.996. A graph derived from this data is shown problem was dramatically reduced by decreasing the in Figure 2. number of PCR cycles from 40 to15. No product was vis- This method was subsequently utilized to assess the ualized when the number of cycles was reduced to 10. proportion of gene-modified hematopoietic cells in It is likely that 15 cycles generate sufficient product for detection by in situ hybridization, but not enough to pro- mote extracellular diffusion. In addition, there is also a reduction in the time available for product leakage. Pri- mers generating a product towards the upper limit of the optimum size range for ISPCR27 were used to reduce extracellular diffusion further. We did not find it neces- sary to use other strategies to obviate this problem, such as the use of multiple primer sets, extra-long products, overlapping primers, post-PCR fixation with paraformal- a b dehyde or biotinylated primers.28,29 A post-hybridization wash of 62°C was used. This is higher than in published Figure 1 (a) Result using indirect ISPCR. Provirus positive and negative protocols but was found to eliminate background with- cells were mixed in equal proportion and subjected to an initial in situ out significantly decreasing detectable product. PCR reaction using unlabeled primers for 15 cycles followed by detection We confirmed our results by performing mixing by in situ hybridization using FITC labeled probes to the amplicon. Posi- experiments using identical cell lines which were either tive cells show strong FITC fluorescence while negative cells show red provirus (neor) positive or negative and compared propidium iodide staining. (b) Negative control using indirect ISPCR. observed with actual percentages (Figure 2). The calcu- Provirus positive and negative cells were mixed in equal proportion and subjected to the method described in (a) except for the omission of Taq lated sensitivity, specificity and correlation coefficients polymerase and primers in separate experiments. No FITC signal is for these experiments were extremely high providing observed. unequivocal validation of the ISPCR approach for differ- Direct quantification of gene transfer with ISPCR C Catzavelos et al 757 Table 1 Scoring data from mixing experiments

Actual ISPCR Positive Total Actual ISPCR Positive Total % positive % positive cells cells % positive % positive cells cells

0 0 0 500 5 7 28 400 0 0 0 100 5 4 17 400 0 0 1 300 5 5 18 400 0 0 0 500 5 11 34 300 0 0 0 1200 5 0 0 300 0 0 0 300 10 10 40 400 0 0 0 500 10 9 211 2311 0 0 0 100 10 18 50 272 0 0 0 300 10 7 21 300 0 0 0 500 10 3 10 300 0 0 0 500 10 7 21 300 0 0 0 500 10 8 33 400 0.0001 0 1 300 20 13 40 300 0.0001 2 5 300 20 8 25 300 0.001 1 2 300 20 14 200 1400 0.001 1 4 300 20 16 49 300 0.01 0 1 300 30 35 80 228 0.01 1 2 300 50 48 103 214 0.1 0 1 300 70 65 205 315 1 5 18 342 90 83 286 343 1 2 12 500 99 98 402 410 1 2 9 500 100 95 477 500 1 0 2 500 100 99 690 700 1 0 0 300 100 96 96 100 1 2 6 300 100 100 100 100 5 17 52 300 100 94 1405 1500 5 7 35 500 100 91 272 300 5 8 40 500 100 97 291 300 5 6 29 500 100 97 389 400 5 12 59 500 100 98 293 300 5 6 24 400 100 97 386 400

Known provirus positive and negative cells were mixed in varying proportions and subjected to ISPCR followed by blinded scoring. The number of positive cells observed was scored as a percentage of the total number counted and compared with the actual percentage of positive cells in the cell mixture. Statistical analysis of this data showed a Pearson correlation coefficient of 0.99. entiating transduced and nontransduced cells, parti- limitations of currently employed technology for target cularly in the 1–5% range. gene detection and could become the method of choice Two recent studies have used DNA in situ PCR to for this assessment. evaluate gene transfer efficiency, both using the direct method. In one of these,30 a true negative control was not achieved and the final results were obtained by subtrac- Materials and methods tion of the number of postive cells in the ‘negative con- trol’ from the number observed in the positive samples. Samples for initial experiments and mixing experiments In the other,31 a single positive and negative control was Jurkat cells with and without stable neor integrant kindly used but the results were not confirmed by any other provided by Patrick Hwu,22,23 were mixed in equal pro- method. portion and slides prepared as described below for the In addition to the stem cell marking trial,24 we are initial experiments to develop the method. Subsequently, currently using this approach in several specific gene mixing experiments were performed with known pro- therapy settings to detect constructs containing MDR1, ␣- portions of neor-positive and neor-negative Jurkatt cells. iduronidase, glucocerebrosidase and adenosine deamin- Several mixing experiments were performed with per- ase genes. In further ongoing studies, we are evaluating centages ranging from 0–100% (see Figure 2). The results the applicability of this approach for assessing gene were analyzed using standard methods for the transfer in tissue sections which will probably be calculation of sensitivity, specificity and coefficient of important in the development of gene therapy for solid correlation.32 tissues. In summary, we have developed a simple, reliable and Clinical samples accurate PCR in situ method for the assessment of gene Patients from the ongoing multiple myeloma gene mark- transfer efficacy. In principle, this approach can be used ing trial received cells transduced with a G1Na super- for any gene delivered in the course of genetic therapy natant (GTI, Gaithersburg, MD, USA) containing the neor for which sequence information is available, obviating marker. At various post-infusion time-points, patient the need to add an exogenous marker gene to the trans- blood and bone marrow samples were obtained and ferred construct. In addition, it addresses many of the assayed for the presence of the transferred gene.25 Direct quantification of gene transfer with ISPCR C Catzavelos et al 758

Figure 2 Comparison of actual and observed numbers of positive cells in mixing experiments. Provirus positive and negative cells were mixed in varying proportion, subjected to ISPCR and scored in a blinded fashion. The observed and actual percentages were compared and subjected to statistical analysis to assess sensitivity, specificity and the coefficient of correlation. The curve generated shows a linear relationship between observed and actual scores with a Pearson correlation coefficient of 0.99. The inset Figures show observed results at 0, 10, 20, 50, 70 and 100% positive cells, respectively.

Peripheral blood and bone marrow samples and subsequent in situ hybridizations were performed Peripheral blood and bone marrow (0.5 ml) was added using the GeneAmp 1000 system (Perkin Elmer Applied to 14 ml of hypotonic KCl (0.075 m) followed by incu- Biosystems Division, Foster City, CA, USA). The neor pri- bation at 37°C for 20 min. Thereafter, the samples were mers used were: forward 5Ј GTG GAG AGG CTA TTC centrifuged at 400 g for 10 min, the supernatant was GGC CTA TGA and reverse 5Ј GTC CAG ATC ATC CTG removed and 3 ml of Carnoy’s fixative (3:1 meth- ATCGAC AAG. The primer annealing temperatures (61.4 anol:acetic acid) was added. The fixation step was and 59.5°C) were calculated using vector NTI 4.0 deluxe repeated three times. (Scientific and Educational Software, Gaithersburg, MD, USA). A 50 ␮l master mix reaction was prepared using T cell and B cell samples 10 mm Tris HCl (pH 8.3), 50 mm KCl, 4.5 mm MgCl2, 0.4 T cell and B cell cultures were established from peri- mm dATP, 0.4 mm dCTP, 0.4 mm dGTP, 0.6 mm dUTP, pheral blood samples in which mononuclear cells were 8 ng neor primer 1, 8 ng neor primer 2, 0.1% BSA, and separated on a Ficoll gradient. T cells were subsequently 10 units of Taq DNA polymerase. The reaction mixture cultured after selection with PHA and IL-2, whereas B was heated to 70°C before the addition of Taq and sub- cells were cultured using EBV.33 Both B and T cell cul- sequently placed on a slide preheated on the GeneAmp tures were harvested using the same method described 1000 in situ PCR system assembly tool. The reaction mix- for blood and bone marrow samples. ture was sealed in contact with the target cells using AmpliCover discs and clips. The slide was subsequently Slide preparation placed in a Perkin-Elmer GeneAmp 1000 in situ PCR ther- Fresh slides were made from bone marrow, peripheral mal cycler and subjected to between 15 and 40 PCR cycles blood, T cell, B cell and various cell line samples by drop- (94°C for 1 min for denaturation, 61°C for 1 min for ␮ × 6 ping 20 l of cell suspension containing 2 10 cells with annealing, and 72°C for 1 min for extension) followed by a transfer pipette on to precleaned slides. These were air a 10 min hold at 72°C. Following PCR amplification, the dried and subsequently dehydrated in a 70%, 85% and slides were washed for 5 min in 2 × SSC (pH 7.0) at 95% ethanol series at room temperature. Slides were room temperature. incubated with RNase (20 mg/ml) for 1 h at 37°C before use. Controls PCR amplification One hundred per cent positive control cells were pre- PCR reagents were obtained from the Perkin Elmer in situ treated with RNAse (20 mg/ml) for 1 h at 37°C before PCR core kit (N808–0197) and all in situ PCR reactions PCR and in situ hybridization. No reduction in signal was Direct quantification of gene transfer with ISPCR C Catzavelos et al 759 observed. Similar slides were pretreated with DNAse I (1 References mg/ml) for 30 min at 37°C followed by the addition of 0.5 m EDTA to stop the reaction. This resulted in the com- 1 Crystal RG. Transfer of genes to humans: early lessons and plete absence of positive signal. obstacles to success. Science 1995; 270: 404–410. 2 Blau HM, Springer ML. Gene therapy – a novel form of drug delivery. New Engl J Med 1995; 333: 1204–1207. Probes 3 Moritz T, Willams DA. Gene transfer into the hematopoietic sys- Two fragments specific for adjacent areas of the neor gene tem. Curr Opin Hematol 1994; 1: 423–428. were isolated from 10 ␮gofaneor-containing plasmid 4 Bordignon C et al. Gene therapy in peripheral blood lympho- (kindly provided by Gary McGarrity, Gene Therapy Inc, cytes and bone marrow for ADA-immunodeficient patients. Gaithersburg, MD, USA), cut in a triple digest with EcoRI, Science 1995; 270: 470–475. 5 Gussoni E et al. Normal dystrophin transcripts detected in PstI and SphI. The EcoRl–Pst digest provided a 235 bp Duchenne muscular dystrophy patients after myoblast trans- fragment near the N-terminus, whereas the PstI/SphI plantation. Nature 1992; 356: 435–438. digest yielded a 352 bp fragment within the mid-section 6 Knowles MR et al. A controlled study of adenoviral-vector- of the gene. The two fragments separated on a 3% low mediated gene transfer in the nasal epithelium of patients with melting point gel, purified and resuspended in 10 mm cystic fibrosis. New Engl J Med 1995; 333: 823–831. Tris HCI (pH 8.0). Both fragments were labeled by ran- 7 Gussoni E et al. A method to detect introduced genes and their dom primers using the Fluor-it Fluorescence Labeling Kit products in gene therapy protocols. Nature Biotech 1996; 14: (Stratagene, La Jolla, CA, USA; 300 383). DNA from each 1012–1016. fragment was labeled by adding random primers, denat- 8 Price J, Turner D, Cepko C. Lineage analysis in the vertebrate uring the sample at 95°C for 5 min followed by the nervous system by retrovirus-mediated gene transfer. Proc Natl − Acad Sci USA 1987; 84: 156–160. addition of Klenow (exo ) and reaction buffer containing 9 Zolotukhin S et al. A ‘humanized’ green fluorescent protein fluorescein-dUTP. The reaction mixture was incubated at cDNA adapted for high level expression in mammalian cells. 37°C for 25 min followed by the addition of a stop J Virol 1996; 70: 4646–4654. reagent before resuspension in water. The final probe 10 Apperley JF, Luskey BD, Williams DA. Retroviral gene transfer concentration was 0.2 ng/ml. of human adenosine deaminase in murine hematopoietic cells: effect of selectable marker sequences on long-term expression. Blood 1991; 78: 310–317. In situ hybridization 11 Riddell SR et al. T cell mediated rejection of gene-modified HIV- Following the PCR amplification and a post-amplification specific cytotoxic T lymphocytes in HIV-infected patients. Nat- wash, 50 ␮l of labeled probe was denatured for 5 min at ure Med 1996; 2: 216–222. 95°C, placed on the slide and sealed using AmpliCover 12 Brenner M. Gene marking. Hum Gene Ther 1996; 7: 1927–1936. 13 Bagasra O et al. Detection of human immunodeficiency virus discs and clips. The GeneAmp in situ PCR thermal cycler type 1 provirus in mononuclear cells by in situ polymerase chain was used for denaturation (5 min at 94°C) and hybridiza- ° reaction. New Engl J Med 1992; 326: 1385–1391. tion (16 h at 37 C). After hybridization, the slides were 14 Embretson J et al. Massive covert infection of helper T lympho- washed for 10 min in a 2 × SSC (pH 7.0) solution at 62°C, cytes and macrophages by HIV during the incubation period of followed by a 5 min wash in 1 × PN (pH 8.0) at room AIDS. Nature 1993; 362: 359–362. temperature. Thereafter, the cells were counterstained 15 Boshoff C et al. Kaposi’s sarcoma-associated herpes virus infects using 10 ␮l of 0.6 ␮g/ml propidium iodide and cover- endothelial and spindle cells. Nature Med 1995; 1: 1274–1278. slipped. 16 Embleton MJ et al. In cell PCR from mRNA: amplifying and linking the immunoglobulin heavy and light V genes within single cells. Nucleic Acids Res 1992; 20: 3831–3837. Visualization and scoring 17 Tokusashi Y et al. Differentiation of the normal and mutant rat Cells were visualized using a triple band pass filter on a albumin genes in hepatic tissue sections by in situ PCR. Nucleic Zeiss Axioscope fluorescence microscope (Carl Zeiss, Acids Res 1995; 23: 3790–3791. 18 Sallstrom J, Zehbe I, Alemi M, Wilander E. Pitfalls of in situ North York, Ontario, Canada). The nuclei of negative polymerase chain reaction (PCR) using direct incorporation of cells stained red and positive cells appeared yellow- labelled nucleotides. Anticancer Res 1993; 13: 1153. green. Three hundred cells were counted in at least four 19 O’Leary JJ et al. In situ PCR: pathologists dream or nightmare? different regions of the slide. Only nonoverlapping cells J Pathol 1996; 178: 11–20. that had good morphology and uniform staining were 20 Teo I, Shaunak S. Polymerase chain reaction in situ: an appraisal scored. of an emerging technique (review). Histochem J 1995; 27: 647–659. 21 Teo I, Shaunak S. PCR in situ: aspects which reduce amplifi- cation and generate false-positive results. Histochem J 1995; 27: Images 660–669. Images were collected using a CCD camera and the 22 Morgan RA et al. Retroviral vectors containing putative internal INSIGHT-IQ Image Processing and Quantitation Com- ribosome entry sites: development of a polycistronic gene trans- puter System (Meridian Instruments, Okemos, MI, USA). fer system and applications to human gene therapy. Nucleic Composites were made using Adobe Photoshop 3.0 and Acids Res 1992; 20: 1293–1299. 23 Lam JS et al. Improved gene transfer into human lymphocytes Power Point 4.0 on a Macintosh computer. using retroviruses with the gibbon ape leukemia virus envelope. Hum Gene Ther 1996; 7: 1415–1422. 24 Stewart AK, Dube I, Kamel-Reid S, Keating A. A phase I study Note added in proof of autologous bone marrow transplantation with stem cell gene During the preparation of this manuscript, the method marking in multiple myeloma. Hum Gene Ther 1995; 6: 107–119. described was used on a series of clinical samples with- 25 Stewart AK et al. Engraftment of gene marked long-term mar- out RNase pretreatment. A false positive error rate of 4% row culture cells (LTMC) in myeloma patients. Blood 1996; 88 was obtained which indicates the importance of this step. (Suppl. 1): 270a (Abstr. 1068). Direct quantification of gene transfer with ISPCR C Catzavelos et al 760 26 Komminoth P, Long AA, Ray R, Wolfe HJ. In situ polymerase 30 Hanania E et al. Results of MDR-1 vector modification trial indi- chain reaction detection of viral DNA, single-copy genes, and cate that granulocyte/macrophage colony-forming unit cells do gene rearrangements in cell suspensions and cytospins. Diag not contribute to post-transplant hematopoietic recovery follow- Mol Pathol 1992; 1: 85–97. ing intensive systemic therapy. Proc Natl Acad Sci USA 1996; 93: 27 Embretson J et al. Analysis of human immunodeficiency virus- 15346–15351. infected tissues by amplification and in situ hybridization 31 Gnatenko D et al. Characterization of recombinant adeno-asso- reveals latent and permissive infections at single-cell resolution. ciated virus-2 as a vehicle for gene delivery and expression into Proc Natl Acad Sci USA 1993; 90: 357–361. vascular cells. J Invest Med 1997; 45: 87–98. 28 Patterson BK et al. Detection of HIV-1 DNA and messenger RNA 32 SAS/SAT User’s Guide (vol 2, 4th edn). The SAS Institute: Cary, in cells by PCR-driven in situ hybridization and flow cytometry. NC, 1989. Science 1993; 260: 976–979. 33 Kinchington D, Berrie E. Separation and maintenance of pri- 29 Chiu K et al. Intracellular amplification of proviral DNA in mary T and B lymphocytes. In: Pollard JW, Walker JM (eds). tissue sections using the polymerase chain reaction. J Histochem Animal Cell Culture, vol 5. Humana: New Jersey, 1989, pp 39–40. Cytochem 1992; 40: 333–341.